Frequently asked questions

What is CriticCode?

CriticCode is an interview-prep workspace. You author a challenge (a real one, the kind your candidate would actually see in the role) and a handful of structured prompts. The candidate works through it in their own words, with an in-page AI collaborator they can brainstorm with — exactly the way they'd use AI on the job. When they submit, you get their answers, the full chat transcript, and every piece of text they pasted in.

Why does this exist?

Most interview formats are trying to filter out AI. That framing is already out of date — AI is part of how engineers work now. We built CriticCode the other way: let the candidate use it, and capture how they use it.

What you get back isn't a score, it's a prep artefact. You walk into the interview already knowing what they assumed, what they asked the AI, what they pushed back on, and what they copy-pasted. The interview itself starts from a real conversation instead of a cold prompt.

How does it work?

You author a challenge in your dashboard (a statement plus whichever prompts matter — assumptions, trade-offs, edge cases, testing strategy, anything). Reuse it across as many candidates as you want. When you're ready, enter the candidate's details, pay via Stripe, and they get an email with a one-time link.

They work through it at their own pace. When they submit, you can review the whole thing at /invitations/:id/submission: every prompt response, the full AI transcript, and highlighted paste events.

How long does it take the candidate?

As long as they want, within seven days. The challenge is open-ended, so there's no single "right time" — but the format is designed for around half an hour of focused work, not a weekend-long take-home.

What makes a good challenge?

Something close to real work. Not a gotcha, not an algorithm puzzle — a scenario your team has actually faced or something similar enough that their thinking transfers. The more ambiguous the challenge, the more signal you get from how they frame it.

Your prompts should surface the thinking you'd want to hear in the interview anyway. Good examples: "Assumptions you'd check before starting", "Trade-offs you considered", "How you'd test this under load", "What you'd do differently at 10× scale".

What languages and stacks are supported?

You write the challenge, so CriticCode is language- and stack-agnostic. Bring a challenge that reflects what the candidate would actually do day to day.

Can I reuse a challenge across candidates?

Yes. Challenges are reusable by design — author it once, send it to as many candidates as you like. Delete it when it's no longer relevant; existing invitations that used it keep their original prompts.

Why let candidates use AI during the assessment?

Because they'll use it on the job anyway. Testing them in an artificially AI-free sandbox tells you almost nothing about how they'll actually work. What matters is how they use it: what they ask, what they keep, what they reject, and how clearly they can explain the final thinking in their own words.

The transcript is the artefact. Read it the same way you'd read a code review — not for "did they ask for help" but for "did their questions get sharper" and "did they accept a bad suggestion without noticing".

What does the AI actually do?

It's a Claude-based collaborator that knows the challenge context and helps the candidate brainstorm, challenge their ideas, or work through details with them. It's not a grader — it has no notion of a "right answer". One thread per invitation, full message responses (no streaming in the MVP).

What if someone pastes the challenge into a different AI?

They can, and we record everything they paste back. The submission view highlights pasted segments in context, so you can see exactly which parts of their answer came from somewhere else. Follow up in the interview — a candidate who can explain pasted reasoning in their own words is very different from one who can't.

Is there a score?

No. There is no automated score — deliberately. We don't think a number captures what you're actually trying to evaluate. What you get is a structured prep artefact: the answers, the AI transcript, and the paste events. You walk into the interview with a real starting point.

How should I read the submission?

Start with the answers — are they specific, or generic? Do they engage with the real constraints of your challenge, or do they float above it? Then skim the AI transcript: did the candidate use it to sharpen their thinking, or to outsource it?

Highlighted paste events are the third signal. They don't automatically mean "cheating" — they might just be moving text around. But they flag exactly what to probe in the follow-up interview.

Is this fair to candidates?

We think so. It's shorter than most take-homes, there's no live-coding pressure, no environment setup, and no hidden rules — the AI is explicitly part of the workspace. Candidates answer in their own words; the goal is clarity, not polish.

Can a candidate retake?

No. Each invitation link is single-use and expires after seven days. A new invitation is a new purchase. This is intentional — the format captures first-pass thinking, not a rehearsed version.

Do candidates need an account?

No. They open the email link, work the challenge, and submit. Nothing to install, no login to create.

How does pricing work?

A flat €29 per invitation, up to 3 challenges included. No subscription, no per-seat pricing, no expiring credits. Author as many challenges as you want for free; you only pay when you actually send an invite. The AI collaborator is included.

How many challenges can I send per candidate?

Up to 3. The candidate works through them one at a time, with a shared AI collaborator across all of them. Price is the same whether you send 1 or 3.

Still have a question?

Reach us at support@critic.codes. We read everything.